(back to the list of all Panda meeting minutes)

Time

8 am PT

Attendees

Wen Guan Wei Yang Brian Yanny Edward Karavakis Fabio Hernandez Jen Adelman-Mccarthy James Chiang Michelle Gower Mikolaj Kowalik Peter Love  

Regrets

Tim Jenness Richard Dubois

Agenda:

  1. CM news
  2. Panda News
    1. Can US-DAC and IDF users use Panda - Do we need to recruit a few power users to help sorting out the use cases?
    2. Panda operation backup (both Wen and Eddie at CERN/GMT-1 timezone). Do we have basic doc on how to check the health of Panda infrastructure at USDF?
    3. IAM 
    4. OpenSearch
  3. HammerCloud
  4. Rucio news 
  5. CHEP 2024 abstract ?


Notes:

  1. CM news
    1. new version of CM tools. Testing CM client/server 
    2. will test memory increase feature from Panda - it may provide efficiency gain.
    3. Long tail in finishing large tasks - Wen: looking for ways to reduce the tail
    4. No longer see jobs being allocated to weka-occupied cores. Is this because of the usage of srun (not expected) or something else?
  2. Panda news
    1. IAM: security concern (by SLAC cyber team) was likely addressed by the developer. Updating USDF IAM will need to go through an intermediate version - We will finally be able to move the IAM from CERN to SLAC.
    2. OpenSearch: no news. 
    3. Richard and Wen are testing instruction for US-DAC and IDF users to use Panda to submit jobs.
    4. Panda operation info: https://panda.lsst.io/v/DM-43733/admin/panda_k8s_usdf.html
  3. HammerCloud: no news. Peter just came back from vacation
  4. Rucio news:
    1. Jhonathan and Brian suspected that the HTTP 500 error (when in-place registering a large number files to Rucio) has something to do with the DB. Dan Speck can be very helpful. The Panda team also went though some of this type of issues. Tim Noble mentioned that Rucio used a large number of connections to the backend DB (~70)
  5. CHEP 2024 abstract: Wei will write a draft and circulate around.